RussianNLP/repa

Dataset Card for REPA Image Source Dataset Description REPA is a Russian language dataset which consists of 1k user queries categorized into nine types, along with responses from six open-source instruction-finetuned Russian LLMs. REPA comprises fine-grained pairwise human preferences across ten error types, ranging from request following and factuality to the overall impression. Each data instance consists of a query and two LLM responses manually annotated to… See the full description on the dataset page: https://huggingface.co/datasets/RussianNLP/repa.

<iframe src="https://silytics.si/embed/item/f6db1341-a210-47a4-8706-aeddd6719567" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.